Measurement model of brass plated tyre steel cord based on wave feature extraction

Abstract

In the production of Truck and Bus Radial (TBR) vehicle tyres, one of the essential components is the wire that supports the tyre. There are several types of tyre wire, one of which is Brass Plated Tyre Steel Cord (BPTSC), produced by Bekaert Indonesia Company. BPTSC object has a micro-size with a diameter of 0.230 mm and has a wave shape. In checking the quality of steel straps, brass-coated tyres are usually measured manually by experienced experts by measuring instruments to measure the diameter using a micrometre, wave amount, and wavelength using a profile projector. The manual measurement process results in inaccuracy due to fatigue in employees' eyes and low lighting and must be repeated, thus, consuming more time. Technological developments that use computer vision are increasingly widespread. Moreover, from the results of studies in various literature, it is proposed to combine the models obtained to find new models to solve this problem. The objectives of this study were to implement and evaluate an automatic segmentation method for obtaining regions of interest, to propose a BPTSC diameter, wave amount, and wavelength measurement model based on its edge, and to evaluate the proposed model by comparing the results with standard and industrial measurement results. The technique to prepare the brass plated tyre steel cord was done in two ways: image acquisition techniques with enhanced image quality, noise removal, and edge detection. Secondly, ground truth techniques were utilised to find the truth about the stages of the image acquisition process. Finally, sensitivity testing was conducted to find the similarity between the acquired images and the ground truth data using Jaccard, Dice, and Cosine similarity method. From 148 wire samples, the average similarity value was 93% by Jaccard, 96% by Dice, and 91% by the Cosine method. Thus, it can be concluded that the acquisition stage of the brass-coated steel tyre cable with image processing techniques can be carried out. For the subsequent process, the pixel distance and the sliding windows model applied can correctly detect the diameter of the BPTSC properly. The wave amount and wavelength of BPTSC objects in the form of waves were measured using several local minima and maxima approaches. This included maxima of local minima maxima distance, the average of local minima maxima distance, and perpendicular shape to centre distance for measuring wave amounts. While for wavelength measurements, the midpoint of local maxima minima distance and the intersection of local maxima minima with a central line were used. Measurement results were evaluated to determine the accuracy and efficiency of the measurement process compared to standard production values using the accuracy, precision, recall, and Root Mean Square Error (RMSE) test. From the evaluation results of the two methods, the accuracy rate of diameter measurement is 97%, wave rate measurement is 95%, and wavelength measurement is 90%. A new model was formed from the evaluation results that could solve these problems and provide scientific and beneficial contributions to society in general and the companies related to this industry.